Optimization of latent properties extraction and data processing of non-stationary objects on the basis of fuzzy genetic algorithms.

The problem is formulated for providing the quality of identification and optimization during data processing of nonstationary objects on the basis of improving and developing methods to extraction and use latent properties of data by generalization of features of statistical and dynamic models, fuzzy sets and fuzzy logic, genetic algorithms (GA). The methods and algorithms are developed to filtering non-stationary components of random time series (RTS) on the basis of statistical strategy, threshold methods, computing circuits of traditional and fuzzy GA. Results of tasks solutions represent mathematical expressions to evaluate the location level and width of optimum borders of control, minimal mean-squared error of data processing. The method is offered for synthesis and generalization of opportunities of segmentation algorithms, algorithms to extracting data properties and regulation the length of initial and final population of generations, level of location, width of filtration borders for non-stationary components of RTS on the basis of fuzzy rules. The designed program complex are focused on use of built-in services, databases and knowledge bases, traditional and fuzzy GA.